Copilot Studio overview and access

1. What is Copilot Studio?

Microsoft Copilot Studio is a graphical low-code tool for building AI agents. Through an intuitive drag-and-drop interface, you can create conversational AI agents and Agent Flows (automation workflows).

One-line summary: “Copilot Studio is a tool for quickly productizing agents.” Its role is different from an AI engine (Azure Foundry) or system implementation (SI).

📖 Reference: Copilot Studio overview

1.1 Key characteristics

📖 Reference: Copilot Studio application card – Key features

1.2 AI models used

Copilot Studio uses a variety of AI models to provide agent experiences.

• OpenAI GPT series (provided by Azure OpenAI Service)
• Anthropic Claude Sonnet family (can be allowed or blocked by organizational policy)
• xAI Grok family (can be allowed or blocked by organizational policy)

When creating an agent, you can select the default model, and performance and response characteristics vary by model. Microsoft does not use customer data to train foundation models.

📖 Reference:
Check which language models you can use by region

2. How to access Copilot Studio

Copilot Studio is a cloud service that you can access directly from a web browser.

2.1 Differences by access method

Important! The Copilot Studio app inside the Teams app is a legacy feature that is not provided in Korea by default. For agent development, make sure to use the web app.

📖 Reference: VS Code extension for Copilot Studio · Quickstart guide

📖 Reference: Required services and network URLs

3. Access permissions and licenses

3.1 License types

To use Copilot Studio, you need the appropriate license.

📖 Reference: Copilot Studio licensing · Power Platform licensing FAQ

3.2 Permission structure

• Security structure: Entra ID-based authentication, Dataverse security roles, and AAD security group-based access restrictions are applied.
• Environment: Copilot Studio runs on Power Platform environments, and agents are created and managed per environment.

📖 Reference: Assign licenses and manage access

4. Billing structure details (Copilot Credits)

4.1 What are Copilot Credits?

Copilot Credits are the common billing unit for Copilot Studio. Credit consumption varies depending on the type of work an agent performs. Starting in September 2025, the billing unit changed from the previous “messages” unit to “Copilot Credits.”

📖 Reference: Billing rates and management

4.2 Credit consumption by feature

As of April 7, 2026. For details, check the official reference document above.

Composite billing example: Tenant Graph Grounding + Generative Answer combination = 12 credits consumed (10 + 2)

💡 Credit cost reference: PAYG basis: 1 Copilot Credit = $0.01 USD

📖 Reference: Copilot Credits billing rates · Agent usage estimator

4.3 Free usage conditions for M365 Copilot users

When users with an M365 Copilot license use a Copilot Studio agent, under certain conditions they can use it for free without consuming credits.

✅ Conditions for free usage (all must be met):

❌ Cases where free usage does not apply:

• Anonymous or Custom Token authentication
• Cases where the M365 Copilot User Identity cannot be verified
• B2C (customer-facing) scenarios
• Automated execution based on a system account (schedule/trigger-based)
• Autonomous/non-interactive work

4.4 Relationship between Power Automate flows and billing

When you use Power Automate flows in Copilot Studio, the billing criteria change.

Key cautions:

• If you change a Cloud Flow to the Copilot Studio Plan, it leaves PA license evaluation and moves to the Copilot Studio meter.
• In this case, you can use premium connectors without a separate PA Premium license, but they are charged against Copilot Credits.

4.5 Credit monitoring and overage management

5. Differences between Copilot Chat / Agent Builder / Copilot Studio

Microsoft provides a variety of AI and automation tools. “Which tool should we use?” is one of the most common questions in practice.

5.1 Tool comparison table

📖 Reference: Choose between Microsoft 365 Copilot and Copilot Studio · Copy agent to Copilot Studio

5.2 Key decision criteria

“If you only need to find information → Agent Builder;
if you want it to actually perform work (actions) or act directly depending on the situation → Copilot Studio.”

5.3 Extending Agent Builder to Copilot Studio

Agent Builder is a tool for quickly creating simple agents, but when advanced capabilities are needed, you can extend it by using Copy to Copilot Studio.

Advanced capabilities available after copying:

• Enhanced lifecycle management: version control, staged deployment, rollback
• Usage monitoring and analytics: user engagement, query trends, performance insights
• Governance controls: role-based access, DLP policies, compliance
• Environment management: separation of development/test/production environments
• Audit and compliance: audit trails, compliance reporting
• Organizational deployment: publish to Teams app store, company-wide deployment through admin approval

6. Ways to use agents

Agents created with Copilot Studio are used in two main ways.

6.1 1. Standalone agents

These agents converse directly with users in channels such as websites, Teams, and Facebook.

Usage examples:

• Customer support: order tracking, returns processing, automated FAQ answers
• Internal FAQ: HR policy guidance, IT manuals, benefits inquiries
• IT help desk: password resets, VPN connection, equipment requests
• Business automation: approval requests, email sending, data registration

📖 Reference: Intended uses of Copilot Studio

6.2 2. Microsoft 365 Copilot extension

Integrate with M365 Copilot as a plugin/extension agent to add your organization’s specialized knowledge and capabilities to the existing Copilot experience.

Example: An employee asks in Copilot Chat,
    "Tell me about our company's annual leave policy"
    → the connected HR Agent answers based on HR documents in SharePoint

📖 Reference: Extend Microsoft 365 Copilot with agents

6.3 3. Agent-to-agent integration (multi-agent)

Agents can not only be used directly, but also provided as tools to other agents.

[User] → [Integrated business agent]
               ├─ HR-related questions → call [HR Agent]
               ├─ IT-related questions → call [IT Agent]
               └─ Expense-related questions → call [Finance Agent]

Through this architecture, you can manage domain-specific expert agents independently while still providing users with a single integrated entry point.